Page 1
How did Covid hit different parts of the world? The visualizations below give you the story. The top two animated plots show the timeline of the spread of the infection and the development of the Twitter narrative. The line plots breaks down the spread by country over time, and puts the regional severity into perspective.
Row
Map of Covid infection, death and recovery this year
Twitter wordcloud this year. China placed in the front and center.
Row
Covid is growing exponentially in the US, India and Brazil
China has flattened the curve. France and the US are getting sick faster than recovering
Think Covid situations in these countries are similar? Let’s unify the scales
Page 2
These weighted networks are made by taking the top 100 words in the English tweets during the first half of the pandemic (left) and the pandemic as a whole (right). If two words show up in the same tweet, then an edge is made to connect them. We see how the network communities changed between June and now.
Before this fall, “trump” is further away from the center of “covid”, connecting to another community of keywords like “tulsa rally”, “mask”, “social distancing”, and “hospital”, but later on shifted to the front and center. This might be because early on in the pandemic, people were generally okay with the president’s handling of it. However as time went on, more people began to think that he is directly related to or responsible for the current covid situation.
Column
Network on Twitter keywords in Spring and Summer
Column
Network on Twitter keywords the whole year
Page 3
This visualization shows the topics discovered by an LDA model. Interact with the bubbles and words to see which cluster represent what topics.
Comparing cluster 3, 4 with cluster 1, we see that cluster 4 is made of key words such as “fauci”, “science” and “vaccine”, whereas cluster 1 is made of “covid”, “death” and “new cases”. This probably means that PC1 represents two sides of people’s attitudes towards covid. Interestingly, three people are ranked high among the keywords: “donald trump”, “ryanafournier”, “joebiden” and “nicolasmaduro”. Joe Biden shows up in cluster 4 whereas Donald Trump shows up in cluster 1.